¶1 This case presents the question of when the confrontation clause requires testimony from lab analysts who conduct forensic tests on evidence. While the United States Supreme Court has grappled with this issue on multiple occasions, a majority of the Court has not adopted a single theory or test. Accordingly, our decision follows the results of recent Supreme Court decisions and proposes a test for expert witnesses that does not conflict with Supreme Court precedent.
¶2 We examine the plain language of the confrontation right: an accused person has a right to confront “the witnesses against him.”1 Reading these words in light of the founders’ intent, the practice of other jurisdictions, and the trajectory of Supreme Court confrontation clause jurisprudence leads us to adopt a rule that an expert comes within the scope of the confrontation clause if two conditions are satisfied: first, the person must be a “witness” by virtue of making a statement of fact to the tribunal, and second, the person must be a witness “against” the defendant by making a statement that tends to inculpate the accused.
*463¶3 Applying this rule shows that there was no violation of Sione Lui’s rights under the confrontation clause when the trial court admitted the results of deoxyribonucleic acid (DNA) testing on samples taken from Lui and from the crime scene, or when it introduced temperature readings taken from Elaina Boussiacos’s body and the ambient environment. Lui’s right to confront a witness was violated by the admission of Boussiacos’s postmortem toxicology results and several statements from the autopsy report, but the errors were harmless. Accordingly, the Court of Appeals is affirmed.
FACTS AND PROCEDURAL HISTORY
¶4 Lui and Boussiacos had a turbulent relationship, marked by mistrust and infidelity. Although they were engaged and living together by the summer of 2000, Boussiacos was uncertain about their marriage plans and she alternated between wearing and not wearing her engagement ring. Boussiacos eventually discovered proof of an affair Lui was having with a married woman, and together the two women trapped him in a lie. Lui was aware that his relationship with Boussiacos was in trouble. He feared that Boussiacos would not return from a trip in mid-2000 and called a friend distraught and crying at the prospect of losing her.
¶5 Boussiacos told her mother that she no longer planned to marry Lui, and in early 2001, Boussiacos made plans to fly to her mother’s home in California. On Friday, February 2, 2001, the night before her flight, Boussiacos dropped off her son from a previous marriage with the boy’s father. Lui told police that Boussiacos returned to the couple’s home at roughly 10:00 p.m., and the couple watched television. According to Lui’s account, Boussiacos packed for the trip, changed into her nightgown, and went to bed.
¶6 Boussiacos never arrived in California. Her mother contacted Lui to report her missing the following Monday. *464On Friday, police found her car in the parking lot of a health club the couple frequented, located near their home. The police discovered Boussiacos’s body in the trunk. The owner of the health club testified that she first noticed the car parked in the lot Saturday morning, February 3, and that it did not move all week. Police arranged for a bloodhound track shortly after discovering Boussiacos’s body. After smelling a sample of Lui’s clothes, the dog followed a scent trail from the lot where the body was found directly to Lui’s front porch.
¶7 Boussiacos’s friends and family agreed that she paid close attention to her personal appearance, taking great care with her dress and makeup when she went out. Her ex-husband testified that she routinely spent two hours on makeup, hair, and clothes before leaving the house. But when found, she had little makeup on and she was dressed in black sweatpants, torn underwear, and a white T-shirt. Investigators noted that she was wearing tennis shoes, but the laces were tied oddly, on the far sides of each shoe, suggesting that her killer had dressed her after death. In addition, Boussiacos’s luggage was packed in an unusual manner, containing several empty containers of hair product and makeup, two hair dryers, and a bottle of nail polish remover without any nail polish.
¶8 In 2007, detectives reviewing cold cases contacted and interviewed Lui. The State subsequently charged Lui with second degree murder in the death of Boussiacos. At trial, in addition to the evidence described above, the State presented expert testimony from Chief Medical Examiner Dr. Richard Harruff and DNA expert Gina Pineda. Harruff’s testimony related to Boussiacos’s autopsy. While Harruff personally reviews the reports for each of the 1,300 autopsies that his office processes each year, the actual autopsy had been performed by Associate Medical Examiner Dr. Kathy Raven. Harruff was not present for the autopsy, and while he believed that he saw the body after the procedure, he could not be sure. However, Harruff did *465not testify to Raven’s conclusions; and, the report was not introduced into evidence. Instead, he referred to photographs of the victim’s injuries taken during the autopsy to testify that in his opinion, the cause of death was asphyxia by manual strangulation or strangulation with a ligature. Based solely on his experience with strangulation, he offered his opinion that it takes roughly four minutes to die in this manner. Harruff also testified to the position of Boussiacos’s body and the odd manner in which she was dressed. While Harruff relied primarily on photographs for this testimony, he made several statements that were taken from the autopsy report.
¶9 Harruff testified that the body’s temperature at the scene was measured at 38.4 degrees Fahrenheit and that the ambient temperature was 30.5 degrees Fahrenheit. He did not take these measurements himself. Rather, Raven took the temperature measurements and recorded them in personal notes that were not part of the autopsy report but were later obtained in discovery. Based on these two temperature data points, Harruff testified to his opinion that although it was “extremely difficult” to fix an exact time of death, death was possible at any time between the second and seventh of February. 10 Report of Proceedings (RP) at 1354-56, 1398-99.
¶10 Harruff also testified to the conclusions of a toxicology report prepared by analyst Martin Hughes of the Washington State Toxicology Laboratory. Harruff did not perform this test personally or supervise it, and he did not offer his professional opinion about the testing methodology. Instead, he recited the report’s conclusion that no drugs, alcohol, or nicotine were found in Boussiacos’s system.
¶11 The Washington State Patrol Crime Laboratory sent DNA samples obtained from the crime scene to two outside DNA laboratories: Orchid Cellmark and Reliagene Tech*466nologies, a company that Orchid had acquired.2 The samples included cuttings from Boussiacos’s shoelaces as well as a vaginal swab and a vaginal wash of Boussiacos’s body. Pineda, Orchid’s associate director and technical leader, testified about her company’s testing of these samples against DNA taken from Lui, Lui’s son from a previous marriage, and Boussiacos’s ex-husband.
¶12 Pineda did not personally participate in or observe the tests, noting that since assuming her director role, she had “stepped away from the lab,” although she did use the electronic data produced during the testing process to create a DNA profile that reflected “[her] own interpretation and [her] own conclusions . ...” 12 RP at 1484, 1507. She offered a document summarizing the test results, which the trial court admitted solely for illustrative purposes, ruling that Pineda could refer to it during her presentation but that it would not go back to the jury room. State Ex. 136. Pineda testified that based on the results of these tests, she could not eliminate Lui or Lui’s son as a major donor of the male DNA found on the shoelaces. Nor could Boussiacos’s ex-husband be eliminated as a donor. The lab’s testing was unable to detect a male profile from the vaginal swab extract. However, Lui or Lui’s son could not be eliminated as a donor of the DNA found in the vaginal wash.
¶13 Lui objected to Harruff’s and Pineda’s testimony on hearsay and confrontation grounds. The trial court rejected his hearsay argument because ER 703 allows experts to rely on hearsay in forming their opinions. It concluded that there was no confrontation violation because Harruff and Pineda were available for cross-examination. A jury found Lui guilty as charged, and the trial court imposed a standard-range sentence of 200 months of confinement.
¶14 Lui appealed, and the Court of Appeals affirmed in a published opinion. State v. Lui, 153 Wn. App. 304, 325,221 *467P.3d 948 (2009). The Court of Appeals concluded that there was no confrontation violation because the expert witnesses testified to their conclusions and Lui had the opportunity to confront them at trial. Id. The court also held that while both experts testified about the content of reports they did not prepare, the underlying reports were “offered to explain the basis for their opinions” pursuant to ER 703 and therefore were not subject to the confrontation clause. Id. at 322-25.
¶15 We accepted review and heard oral argument, but before issuing a decision, we granted the State’s motion to file supplemental briefs addressing Bullcoming v. New Mexico,_U.S._, 131 S. Ct. 2705, 180 L. Ed. 2d 610 (2011). We then stayed our decision pending the United States Supreme Court’s decision in Williams v. Illinois,_ U.S._, 132 S. Ct. 2221, 183 L. Ed. 2d 89 (2012) (plurality opinion). Following the Supreme Court’s decision in Williams, we lifted the stay and heard reargument in Lui’s case.
ANALYSIS
¶16 As an initial matter, we must decide whether to analyze Lui’s claims solely under the Sixth Amendment to the United States Constitution or separately under article I, section 22 of the Washington Constitution.
I. Article I, Section 22
¶17 Article I, section 22 of the Washington Constitution provides that “[i]n criminal prosecutions the accused shall have the right... to meet the witnesses against him face to face . . . . ” While Lui relies primarily on the Sixth Amendment, he raises article I, section 22 as an alternate ground for relief in the event the court concludes his federal confrontation clause rights were not violated.
¶18 We consider six nonexclusive criteria when determining whether a provision of our state constitution should *468be interpreted independently from its federal analogue: (1) the textual language, (2) differences in the texts, (3) constitutional history, (4) preexisting state law, (5) structural differences, and (6) matters of particular state or local concern. State v. Gunwall, 106 Wn.2d 54, 58, 720 P.2d 808 (1986).
¶19 This court has concluded that article I, section 22 merits an independent analysis as to both the manner and the scope of the confrontation right. State v. Pugh, 167 Wn.2d 825, 835, 225 P.3d 892 (2009). Therefore, a full Gunwall analysis “is no longer necessary.” Id. Rather, we look to “ ‘ “whether the unique characteristics of the state constitutional provision and its prior interpretations actually compel a particular result.” ’ ” Id. (quoting State v. Chenoweth, 160 Wn.2d 454, 463,158 P.3d 595 (2007) (quoting City of Seattle v. McCready, 123 Wn.2d 260, 267, 868 P.2d 134 (1994))). This entails “an examination of the constitutional text, the historical treatment of the interest at stake as reflected in relevant case law and statutes, and the current implications of recognizing or not recognizing an interest.” Chenoweth, 160 Wn.2d at 463. In this case, none of these factors calls for an independent reading of article I, section 22.
¶20 The text of article I, section 22 does not compel a result different from that under the Sixth Amendment. Both the Sixth Amendment and article I, section 22 protect a variety of criminal procedural rights; the relevant right is phrased in the federal constitution as the right “to be confronted with the witnesses against him,” while the state constitution uses the language “to meet the witnesses against him face to face .... ” U.S. Const, amend. VI; Const. art. I, § 22. On the face of these provisions, article I, section 22 is unique in that it uses the language “face to face” where the Sixth Amendment does not. However, in State v. Foster, 135 Wn.2d 441, 462-63, 957 P.2d 712 (1998), a plurality of this court declined to give literal effect to the “face to face” language. We held that “the meaning of the words used in the parallel clauses is substantially the same.” Id. at 459.
*469¶21 But even if we read the “face to face” language literally, it would not affect the resolution of this case. There is no question that Lui confronted the State’s witnesses “face to face”; the question is whether the State presented the correct witnesses. If the analysts who worked on the crime scene samples were “witnesses against” Lui, then neither the state nor the federal confrontation clause was satisfied. If they were not “witnesses against” Lui, then neither the state nor the federal confrontation clause would require the State to produce them at trial. Nothing about the unique language of article I, section 22 compels a particular result here.
¶22 Our prior interpretations of article I, section 22 similarly do not compel a particular result. We have consistently rejected arguments that the state confrontation clause provides greater protection than the federal confrontation clause. See Pugh, 167 Wn.2d at 840-45 (excited utterance hearsay exception does not violate state confrontation rights); State v. Shafer, 156 Wn.2d 381, 391-92, 128 P.3d 87 (2006) (child hearsay statute does not violate state confrontation rights); Foster, 135 Wn.2d at 470 (testimony by closed-circuit television does not violate state confrontation rights). These decisions are consistent with early decisions by Washington courts admitting documentary evidence in lieu of live testimony, as well as prestatehood statutes allowing depositions to be introduced at trial. See Foster, 135 Wn.2d at 462. Lui has not shown any particular Washington tradition protecting the right to confrontation over and above the federal standard.
¶23 Finally, Lui does not brief the current implications of recognizing or not recognizing an expanded confrontation interest under the Washington Constitution, other than that judges and litigants will benefit from knowing which witnesses will appear at trial. But the interest in knowing which witnesses will appear at trial, as well as the consti*470tutional values underlying article I, section 22, are adequately addressed by the Sixth Amendment test we articulate below. Perhaps in another case there may be occasion to recognize a broader state confrontation right, but these facts do not give us reason to do so.
¶24 Neither the constitutional text, the historical treatment of the confrontation right, nor the current implications of adopting a broader confrontation right support an independent reading of article I, section 22 in this case. Accordingly, we analyze Lui’s claim solely under the federal confrontation clause.
II. Confrontation Clause
¶25 United States Supreme Court case law on the confrontation clause is somewhat fragmented and does not provide a controlling rule for cases like Lui’s that involve expert witnesses. As we explain below, in the case of nonexpert witnesses, a majority of the Supreme Court has settled on the primary purpose test as the controlling confrontation clause rule. But in the case of expert witnesses, the members of the Court are divided into two groups of four justices each, with Justice Thomas voting independently based on his unique interpretation of the confrontation clause. In the absence of an authoritative Supreme Court majority rule, we must rely on the plain language of the confrontation clause: an accused person has a right to confront “the witnesses against him.”3 As we explain below, the founders’ intent and the practice of other jurisdictions can help us to interpret these words and ultimately reach a working rule for confrontation of expert witnesses: a person is a “witness” for confrontation clause purposes only if he or she makes some statement of fact to the court (as opposed to merely processing a piece of evidence) and that statement of fact bears some inculpatory character (meaning that the evidence, without the need for *471expert interpretation, bears on some factual issue in the case). Under this test, the court did not violate the confrontation clause when it admitted the DNA evidence through Pineda and the temperature evidence through Harruff. However, the toxicology evidence runs afoul of this test.
A. Supreme Court Confrontation Clause Jurisprudence — A Core Rule, and Uncertainty on the Periphery
¶26 Before 2004, confrontation clause jurisprudence was governed by the indicia of reliability test. See Ohio v. Roberts, 448 U.S. 56, 66, 100 S. Ct. 2531, 65 L. Ed. 2d 597 (1980), abrogated by Crawford v. Washington, 541 U.S. 36, 69, 124 S. Ct. 1354, 158 L. Ed. 2d 177 (2004). In Crawford, the Supreme Court jettisoned the indicia of reliability test. What replaced the indicia of reliability test is less clear: in the years since Crawford, the Court has issued increasingly fractured sets of opinions in five major confrontation clause cases.
¶27 The six cases naturally divide into two sets of three decisions. In Crawford, Davis, and Bryant, the Court dealt with conventional, nonexpert witnesses who had witnessed or had been the victims of the subject crimes. Crawford, 541 U.S. 36; Davis v. Washington, 547 U.S. 813, 126 S. Ct. 2266, 165 L. Ed. 2d 224 (2006); Michigan v. Bryant, 562 U.S. 344, 131 S. Ct. 1143, 179 L. Ed. 2d 93 (2011). The Court was unanimous in Crawford, and a solid majority decided Davis and Bryant. But Justice Thomas articulated the principle that has consistently guided his confrontation clause decisions: “I agree with the Court that the admission of Covington’s out-of-court statements did not violate the confrontation clause, but I reach this conclusion because Covington’s questioning by police lacked sufficient formality and solemnity for his statements to be considered ‘testimonial.’” Bryant, 131 S. Ct. at 1167 (Thomas, J., concurring) (emphasis added).
*472¶28 In the second triad of cases, the Court considered the admissibility of laboratory analysis reports where the analyst who had performed the testing did not testify. Melendez-Diaz v. Massachusetts, 557 U.S. 305, 129 S. Ct. 2527,174 L. Ed. 2d 314 (2009); Bullcoming, 131 S. Ct. 2705; Williams, 132 S. Ct. 2221. In these laboratory analysis cases, Chief Justice Roberts and Justices Kennedy, Breyer, and Alito consistently voted together; Justices Scalia, Ginsburg, Stevens, and Souter (and eventually Justices Sotomayor and Kagan) consistently voted together; and Justice Thomas independently looked for sufficient formality and solemnity. We turn now to both groups of cases.
1. The Core Rule: Conventional Witnesses
¶29 In Crawford, the defendant’s wife told the police that the victim had been unarmed and the prosecution introduced her statements without calling her to the stand. 541 U.S. at 40. The Court noted that the seminal example of an out-of-court testimonial statement is the trial of Sir Walter Raleigh, in which the prosecution read into evidence Lord Cobham’s ex parte statements inculpating Raleigh. Id. at 44. It is this sort of “civil-law mode of criminal procedure, and particularly its use of ex parte examinations as evidence against the accused,” that the confrontation clause is primarily concerned to exclude. Id. at 50. Therefore, Crawford was uncontroversially decided. The Court unanimously held that “Sylvia Crawford’s statement is testimonial under any definition” and thus must be subject to cross-examination notwithstanding reliability. Id. at 61. The Court further held that Sylvia’s statements were not reliable in the first instance because she was “herself a potential suspect” with an incentive to lie because the police asked leading questions and because she did not see the entire altercation. Id. at 66.
¶30 The Court revisited and refined its definition of “testimonial” statements in Davis, 547 U.S. 813. In the first *473of two consolidated cases, Adrian Davis’s former girlfriend called 911 to report that Davis was presently assaulting her. Id. at 817. The 911 recording was admitted into evidence, and Davis was convicted of assault. Id. In the second case, officers arrived at Hershel Hammon’s house after an altercation between him and his wife; one of the officers stayed with Hammon while the other officer interrogated Hammon’s wife in a separate room. Id. at 819-20. Hammon’s wife authored a “ ‘battery affidavit’ ” that the State introduced at trial. Id. at 820.
¶31 In Davis, the Court held that whether a statement was testimonial depended on the declarant’s purpose in making the statement. The Court held that the statements of Davis’s former girlfriend had been made “to enable police assistance to meet an ongoing emergency.” Id. at 822. The Court held that a statement is testimonial when “the primary purpose of the interrogation is to establish or prove past events potentially relevant to later criminal prosecution.” Id. That is, a statement is “testimonial” if it is “solely directed at establishing the facts of a past crime, in order to identify (or provide evidence to convict) the perpetrator.” Id. at 826. Under this analysis, the ex parte statements were properly admitted in Davis’s case and improperly admitted in Hammon’s case. Id. at 834.
¶32 Justice Thomas disagreed, dissenting on the ground that testimony must necessarily bear “some degree of solemnity.” Id. at 836 (Thomas, J., concurring in part and dissenting in part). Therefore, affidavits, depositions, prior testimony, and other statements obtained through a “formalized process” fell within the scope of the confrontation clause, but an informal talk with the police would not. Id. at 836-37. For that reason, Thomas argued, the statements in both Davis’s and Hammon’s cases were admissible. Id. at 834.
¶33 The Crawford consensus began to unravel in Bryant, 131 S. Ct. 1143. There, the trial court admitted into evidence the statement of a gunshot victim who identified *474the shooter to responding police officers. 131 S. Ct. at 1150. Six justices of the Court held that, as in Davis, the statements did not constitute an “out-of-court substitute for trial testimony,” id. at 1155, because the declarant’s statements were not meant for use at trial but rather to help resolve the ongoing emergency of “ ‘an unknown shooter who remains at large ....’” Id. at 1158 (quoting amicus brief). The nature of the emergency diminished the “prospect of fabrication,” thus excusing the statements from the confrontation clause. Id. at 1157.
¶34 Justice Scalia dissented, arguing that the primary purpose of the statements was to provide evidence against the defendant. From the declarant’s perspective, the emergency had already ended. Id. at 1171 (Scalia, J., dissenting). Therefore, “his statements had little value except to ensure the arrest and eventual prosecution of Richard Bryant.” Id. at 1170. Justice Ginsburg also dissented, agreeing in full with Justice Scalia’s analysis. Id. at 1176-77 (Ginsburg, J., dissenting).
¶35 Justice Thomas again broke from the primary-purpose test altogether. In concurrence, he argued that the test should be whether the statements were formal and solemn. Id. at 1167 (Thomas, J., concurring).
2. Scientific Evidence: Three Perspectives
¶36 In the next three decisions, the justices divided into three camps as the Court turned from examining statements by conventional witnesses to examining laboratory analysis reports. In Melendez-Diaz, 557 U.S. 305, Justices Scalia, Stevens, Souter, and Ginsburg, joined by Justice Thomas, held that three certificates identifying bags of powder as “cocaine” were testimonial, as the certificates were functionally equivalent to affidavits and were created for the primary purpose of providing evidence for trial. Id. at 310-11. In fact, “the sole purpose of the affidavits was to provide ‘prima facie evidence of the composition, quality, and the net weight’ of the analyzed substance,” necessary *475elements of the crime under Massachusetts law. Id. at 311 (quoting Mass. Gen. Laws ch. 111, § 13). These sworn statements of fact were admitted into evidence and were “functionally identical to live, in-court testimony, doing ‘precisely what a witness does on direct examination.’ ” Id. at 307, 310-11 (quoting Davis, 547 U.S. at 830). Because the witnesses were not subject to cross-examination, the admission of the certificates violated the confrontation clause. Id. at 308.
¶37 Justice Thomas wrote separately to reaffirm that his support for the majority was conditioned on the formal nature of the affidavits at issue. Id. at 329-30 (Thomas, J., concurring).4 And indeed, the majority’s definition of testimony subject to the confrontation clause was qualified by a requirement of formality such as former testimony, statements under oath, or other indicia of formality. Id. at 310.
¶38 Justice Kennedy, joined by Chief Justice Roberts and Justices Breyer and Alito, dissented on the ground that laboratory analysts were not “ ‘witnesses against’ ” a defendant, as they did not bear “personal knowledge of some aspect of the defendant’s guilt.” Id. at 343-44 (Kennedy, J., dissenting). Justice Kennedy offered three distinctions between laboratory analysts and conventional witnesses: “a conventional witness recalls events observed in the past, while an analyst’s report contains near-contemporaneous observations of the test”; “an analyst observes neither the crime nor any human action related to it”; and “laboratory tests are conducted according to scientific protocols; they are not dependent upon or controlled by interrogation of any sort.” Id. at 345-46.
¶39 In Bullcoming, 131 S. Ct. 2705, the state introduced a certificate recording the defendant’s blood alcohol level at 0.21 grams per hundred milliliters through a coworker of *476the laboratory analyst who had not observed or reviewed the actual testing. Id. at 2710-12. Again, the Court declared the evidence inadmissible by similar divisions as in Melendez-Diaz: Justices Scalia and Ginsburg, now joined by Justices Sotomayor and Kagan (who had replaced Justices Stevens and Souter), and Justice Thomas joining in part.5 The Court drew parallels to Melendez-Diaz, noting that the certificate had an “ ‘evidentiary purpose,’ ” that it was created “in aid of a police investigation,” and that it was formalized. Id. at 2717 (quoting Melendez-Diaz, 129 S. Ct. at 2532). Therefore, the certificate was testimonial, which left the Court to determine whether the State had satisfied its confrontation clause burden. It had not; the witness had not participated in the test and could not speak to the procedures used or observations made. Id. at 2713. The witness had no function except as a “surrogate,” merely relaying the conclusions of another. Id. at 2715.
¶40 Justice Sotomayor wrote separately to emphasize the limited reach of Bullcoming, articulating the factual limits of the case: the sole purpose of the certificate was to be introduced into evidence; the witness who testified at trial was not a “supervisor, reviewer, or someone else with a personal, albeit limited, connection to the scientific test at issue”; the testifying witness did not give “his independent opinion about underlying testimonial reports that were not themselves admitted into evidence”; and the document introduced by the State was not limited to “only machine-generated results.” Id. at 2722 (Sotomayor, J., concurring). Chief Justice Roberts and Justices Kennedy, Breyer, and Alito again dissented, arguing that the report was “impartial” and “prepared by experienced technicians in laboratories that follow professional norms and scientific protocols.” Id. at 2726 (Kennedy, J., dissenting).
*477¶41 Thus, in Melendez-Diaz and Bullcoming, the four-judge block of Justices Scalia, Ginsburg, Sotomayor, and Kagan was joined by Justice Thomas to find a confrontation clause violation. But in the next case, Williams, 132 S. Ct. 2221, Justice Thomas joined the four-judge plurality of Chief Justice Roberts and Justices Kennedy, Breyer, and Alito to find no confrontation clause violation. The issue was whether “Crawford bar [red] an expert from expressing an opinion based on facts about a case that have been made known to the expert but about which the expert is not competent to testify.” Id. at 2227. In Williams, an expert testified that a DNA profile taken from a rape victim matched a DNA profile recovered from the defendant. Id. at 2230. The expert did not prepare the DNA profile; rather, she relied on a DNA profile prepared by an outside laboratory. Id. at 2229. No one from that laboratory was subject to cross-examination. See id. at 2227,2230. Justice Alito wrote for the four-judge plurality, including Chief Justice Roberts and Justices Kennedy and Breyer, offering “two independent reasons” for finding no violation of the confrontation clause. Id. at 2244. First, the expert’s reliance on the previous steps in the DNA analysis was not offered to prove the truth of the matter asserted. Id. at 2228. As a “second, independent basis” for the decision, Justice Alito pointed out that the DNA profile was produced before the defendant was identified as a suspect and “the profile that Cellmark provided was not inherently inculpatory.” Id.
¶42 The four justices who had voted together in Bull-coming — Justices Scalia, Ginsburg, Sotomayor, and Kagan— again voted together in Williams. This time Justice Kagan wrote for the four justices. Justice Kagan saw nothing wrong with the expert witness’s testimony that two DNA profiles matched each other, for this was “a straightforward application” of her expertise. Id. at 2270 (Kagan, J., dissenting). Rather, the Court split on the provenance of the victim’s DNA profile, that is, whether the expert affirmed that one of the profiles she was comparing had actually *478come from the victim, without having participated in creating that profile. Id. at 2236. Justice Kagan opined that the expert’s testimony required the jury to accept the validity of a DNA test that had not been scrutinized by cross-examination. Id. at 2268-69 (Kagan, J., dissenting).
¶43 As in Melendez-Diaz and Bullcoming, Justice Thomas provided the decisive fifth vote, but in Williams he concluded that the DNA lab reports lacked sufficient formality or solemnity to be considered testimonial. Id. at 2260-61 (Thomas, J., concurring in judgment). And none of these three cases provide a single clear rule because Justice Thomas provided the fifth critical vote in all three cases based on his individual theory that evidence is testimonial only if it bears indicia of formality and solemnity.
¶44 The dissent accuses us of counting perspectives and camps rather than signatures. Dissent at 500. However, counting signatures ignores the fact that a majority of the Court has never agreed on a test for expert witnesses, making it very difficult for courts to effectively follow. Four justices joined an opinion holding that the confrontation clause does not apply to expert witnesses when expressing their own conclusions, four justices attached no importance to the fact that evidence came in through an expert witness, and one justice focused on the solemnity of the evidence relied on by an expert witness. Williams, 132 S. Ct. at 2228 (plurality opinion), 2260 (Thomas, J., concurring in judgment), 2269-70 (Kagan, J., dissenting). Even if we count signatures, our decision is consistent with the five justices in Williams who agree that experts may rely on and disclose independent DNA laboratory results when testifying about their own conclusions without violating a defendant’s confrontation rights. Id. at 2240 (plurality opinion), 2255 (Thomas, J., concurring in judgment). Our test respects the five justices in Williams while recognizing that in some circumstances an expert witness’s testimony may trigger the confrontation clause. Our opinion also does not disregard the results in Melendez-Diaz and Bullcoming *479because we address only statements made by expert witnesses and not formalized certificates that are the equivalent of affidavits.
¶45 In addition to there being no clear reasoning for expert witnesses, no ruling of the Court is directly on point here. In three important ways, this case brings us into uncharted constitutional territory. First, Melendez-Diaz did not reach back to encompass every factual predicate behind an expert witness’s findings. The confrontation clause does not demand the live testimony of “anyone whose testimony may be relevant in establishing the chain of custody, authenticity of the sample, or accuracy of the testing device .... ” Melendez-Diaz, 557 U.S. at 311 n.l. In other words, while a break in the chain of custody might detract from the credibility of an expert analysis of some piece of evidence, this break in the chain does not violate the confrontation clause. Id. Second, Bullcoming expressly did not reach the confrontation clause status of raw data generated by an automated process without human input. Rather, the subject matter of the confrontation clause concerns those “past events and human actions not revealed in raw, machine-produced data . ...” 131 S. Ct. at 2714 (emphasis added); see also id. at 2723 (Sotomayor, J., concurring) (“This is not a case in which the State introduced only machine-generated results, such as a printout from a gas chromatograph. . . . Thus, we do not decide whether ... a State could introduce ... raw data generated by a machine in conjunction with the testimony of an expert witness.”). Finally, Williams did not address how the confrontation clause applies to the “panoply of crime laboratory reports and underlying technical statements written by (or otherwise made by) laboratory technicians.” 132 S. Ct. at 2244-45 (Breyer, J., concurring). The same question Williams did not reach — the confrontation clause status of forensic reports, expert witnesses, and the technical data underlying their conclusions — is now squarely before us.
*480B. “Witnesses Against” — Reaching a Test
¶46 In the absence of binding Supreme Court precedent for a rule, we now turn to the plain language of the confrontation right. By its own terms, the confrontation right applies only to “the witnesses against [the defendant] .”6 As we explain below, the word “witness” indicates the act of attesting to facts, while the word “against” indicates that the facts attested to must be adversarial in nature.
¶47 The act of imparting factual information to the court is the sine qua non of a witness. Crawford tells us that a “witness” is a person who “ ‘bear[s] testimony’ ” and that “testimony” is “ ‘[a] solemn declaration or affirmation made for the purpose of establishing or proving some fact.’ ” 541 U.S. at 51 (alteration in original) (quoting 2 Noah Webster, An American Dictionary of the English Language (1828)). This definition does not sweep in analysts whose only role is to operate a machine or add a reagent to a mixture.
¶48 Justice Kagan pointed out in her dissent in Williams that cross-examining a witness could be valuable in order to reveal erroneous lab work. 132 S. Ct. at 2264-65 (Kagan, J., dissenting) (“Hence the genius of an 18th-century device as applied to 21st-century evidence: Cross-examination of the analyst is especially likely to reveal whether vials have been switched, samples contaminated, tests incompetently run, or results inaccurately recorded.”). The live testimony of a subordinate analyst may be desirable, but the question is whether it is constitutionally required — and as the Court recognized in Melendez-Diaz, the potential to introduce error does not a “witness” make. In Melendez-Diaz, Justice Kennedy pointed out that many people might be involved in a single drug test and that each of those people had the “power to introduce error,” and he asserted that requiring *481all or “even one of these individuals to testify threatens to disrupt if not end many prosecutions . . .. ” 557 U.S. at 333 (Kennedy, J., dissenting). In response, Justice Scalia clarified that the confrontation clause does not demand the live testimony of “anyone whose testimony may be relevant in establishing the chain of custody, authenticity of the sample, or accuracy of the testing device____” Id. at 311 n.l. Although the prosecution is indeed obliged to establish the chain of custody, “this does not mean that everyone who laid hands on the evidence must he called.”Id. (emphasis added). Rather, gaps in the chain of custody go to the weight of the evidence and not its admissibility. Id. (citing United States v. Lott, 854 F.2d 244, 250 (7th Cir. 1988)).
¶49 In other words, merely laying hands on evidence, DNA or otherwise, does not a “witness” make — something more is required. In Melendez-Diaz, an analyst became a witness by preparing a statement affirming that a substance was cocaine. Id. at 311. In Bullcoming, an analyst became a witness by giving live testimony that the defendant’s blood alcohol level was 0.21. 131 S. Ct. at 2713. Our analysis here is no different: we are interested in experts who make statements to the court, not people who “la[y] hands on the evidence .... ” Melendez-Diaz, 557 U.S. at 311 n.1.
¶50 Not everyone who makes some affirmation of fact to the tribunal will fall under the confrontation clause. The word “against” implies some adversarial element— some capacity to inculpate the defendant.7 As eight justices (six current) of the Court tell us, the central object of the confrontation clause is “interrogations solely directed at establishing the facts of a past crime, in order to identify (or provide evidence to convict) the perpetrator.” Davis, 547 U.S. at 826 (emphasis added). The New York Court of Appeals has recognized the necessity for a testimonial statement to *482bear some inculpatory character; among other factors, New York looks to “ ‘whether the report accuses the defendant by directly linking him or her to the crime.’ ” People v. Pealer, 20 N.Y.3d 447,454, 985 N.E.2d 903,962 N.Y.S.2d 592 (2013) (quoting People v. Brown, 13 N.Y.3d 332, 339-40,918 N.E.2d 927, 890 N.Y.S.2d 415 (2009)).
¶51 Reading the words “witness” and “against” together, in the context of Supreme Court hints and the reasoned practices of other jurisdictions, gives us a working rule. If the declarant makes a factual statement to the tribunal, then he or she is a witness. If the witness’s statements help to identify or inculpate the defendant, then the witness is a “witness against” the defendant.8
¶52 The dissent misunderstands our rule and holding and misstates Supreme Court precedent. This is evident in six ways.
¶53 First, the dissent misstates the effect of this opinion, fretting that we have concluded that “laboratory reports are not testimonial” and that “a supervisor can recite the testimony of a subordinate.” Dissent at 498. On the contrary, our holding reaches only so far as expert witnesses and identifies when the confrontation clause is satisfied by their cross-examination. Today’s opinion does not allow laboratory reports to be admitted into evidence *483and used against a defendant without effective cross-examination. Nor does it allow a laboratory supervisor to parrot the conclusions of his or her subordinates. Instead, our test allows expert witnesses to rely on technical data prepared by others when reaching their own conclusions, without requiring each laboratory technician to take the witness stand. The test does nothing more.
¶54 Second, the dissent must acknowledge that the United States Supreme Court would reach the same result as this opinion under these facts. The result in Williams was that a forensic specialist was permitted to rely on an outside laboratory’s DNA profile when testifying that it matched a sample of the defendant’s blood without violating the defendant’s confrontation rights. See Williams, 132 S. Ct. at 2228 (plurality opinion), 2255 (Thomas, J., concurring in judgment). Our opinion reaches the same result today: experts may rely on DNA profiles created by other laboratory analysts when concluding there is a DNA match without violating the confrontation clause.
¶55 Third, the dissent misleadingly claims that the United States Supreme Court rejected our test on multiple occasions. Dissent at 502-03. The holdings the dissent relies on do not conflict with, or preclude, our test. In Crawford, the Court held that playing a spouse’s tape-recorded statements during her husband’s criminal trial violated his confrontation rights when he was not given an opportunity to cross-examine her. 541 U.S. at 40, 68-69. Crawford does not preclude our test because it does not address expert witnesses. In Melendez-Diaz, the Court held that there was a violation of a defendant’s confrontation rights when the court admitted affidavits reporting the results of forensic analysis, stating that seized material was cocaine without the witness being subject to cross-examination. 557 U.S. at 307, 329. Melendez-Diaz does not pre*484elude our test because the holdings are consistent. If a statement is used against a defendant in court, the declarant must be subject to cross-examination. In Bullcoming, the Supreme Court held that it violated the confrontation clause to “introduce a forensic laboratory report containing a testimonial certification, made in order to prove a fact at a criminal trial, through the in-court testimony of an analyst who did not sign the certification or personally perform or observe the performance of the test reported in the certification.” 131 S. Ct. at 2713. Bullcoming does not conflict with the test for expert witnesses. An expert witness may not parrot the conclusions of others and circumvent the confrontation clause, as is seen later in our analysis of the toxicology results. An expert may, however, rely on the work of laboratory technicians when reaching his or her conclusion. Therefore, the holdings of these cases do not preclude the confrontation clause test for expert witnesses.
¶56 Fourth, contrary to the claims of the dissent, we agree that if DNA evidence, or other scientific or technical evidence, is used against a defendant in court, the confrontation clause is implicated. The cases cited by the dissent that discuss “neutral” witnesses do so in response to arguments that a witness is exempt from the confrontation clause when the statements are not used against the defendant.9 See Melendez-Diaz, 557 U.S. at 313-14. The Court has repeatedly recognized a witness is not excluded *485from the confrontation clause requirements because his or her testimony appears to be neutral. See id. Accordingly, DNA evidence is not exempt from the confrontation clause because it is neutral. Our test requires cross-examination, but only cross-examination of the witness who gives meaning to raw data. Not every laboratory analyst is required to testify. See Williams, 132 S. Ct. at 2228 (plurality). If DNA evidence is used in trial, someone must be subject to cross-examination. The “someone” required by the confrontation clause is the person who has made the final comparison that is used against the defendant.
¶57 Fifth, the dissent insists that we are wishing away Melendez-Diaz and Bullcoming. Dissent at 503. In fact, we too recognize the holdings of these cases. However, there are distinctions between the facts and legal holdings in those cases and the case before us today. The dissent fails to grasp these distinctions. The Supreme Court has never clearly set forth the confrontation clause requirements for when an expert witness relies on the work of others to arrive at his' or her own conclusion.
¶58 Sixth, the dissent attaches great importance to Melendez-Diaz and the fact that Justice Thomas signed the majority opinion. But the dissent ignores the fact that Justice Thomas also concurred, spelling out his reason for concurring — the documents admitted in Melendez-Diaz possessed the requisite indicia of formality. In any event, the holding in the Melendez-Diaz opinion was narrower than the dissent would have us believe. After discussing various forms of “ ‘core class’ ” testimony, the Melendez-Diaz majority concluded that there was “little doubt that the documents at issue [fell] within the ‘core class of testimonial statements’ thus described.” 557 U.S. at 310 (quoting Crawford, 541 U.S. at 51). The statements were the equivalent of affidavits, which fit Justice Thomas’s view of the confrontation clause applying only to formalized testimony. See id. at 329 (Thomas, J., concurring). The Melendez-Diaz holding does not extend beyond the equivalent of affidavits or other formalized testimony.
*486¶59 Nothing in prior Supreme Court decisions precludes our test. Applying the test shows that Orchid’s DNA analysts were not witnesses against the defendant, but rather facilitated Pineda’s role as an expert witness. Similarly, taking the temperature of Boussiacos’s body was not an inculpatory act; it helped the expert estimate the time of death. Finally, while the toxicology report and Harruff’s statements taken from the autopsy report helped to inculpate Lui, their admission was harmless error.
C. DNA Evidence
¶60 Lui argues that the State violated the confrontation clause when it introduced DNA evidence through a supervisor, Pineda, rather than the analysts who physically conducted the DNA testing. If Pineda had relayed the observations or memory of conventional witnesses, then Lui would be correct. But DNA evidence differs in several important ways from the testimony of conventional witnesses. As we explain below, the DNA testing process does not become inculpatory and invoke the confrontation clause until the final step, where a human analyst must use his or her expertise to interpret the machine readings and create a profile. Pineda used her expertise to create a factual profile that incriminated Lui, and therefore Pineda was the appropriate witness to introduce the DNA evidence.
1. Background — The DNA Testing Process
¶61 For the first step in DNA analysis, the analyst takes a sample from the evidence recovered from the crime scene. For example, here, an analyst took a clipping from Boussiacos’s shoelaces for further analysis. In no sense does an analyst become a “witness” by extracting a sample for testing; the act is not testimonial because no statement has been made yet. This preliminary step in the analysis is essentially part of the chain of custody: an error at this stage goes to its weight and not its admissibility. See Melendez-Diaz, 557 U.S. at 311 n.1 (citing Lott, 854 F.2d at 250).
*487¶62 Second, the analyst uses chemicals to break down the sample and release DNA molecules. Here, the chemical compounds, not the human analyst, do all the work. To reiterate, no authority states that the analyst becomes a “witness against” anyone by virtue of adding chemicals to a mixture. Rather, while the analyst’s testimony might be helpful in bolstering the authenticity of the sample or the accuracy of the machine, Melendez-Diaz does not require that testimony. Id.
¶63 Third, the analyst measures the amount of DNA recovered in the second step and replicates the DNA through a process called polymerase chain reaction, which a witness in this case described as “chemically photocopying the 13 different areas of DNA . ...” 9 RP at 1155. This is done in order to amplify a small amount of DNA to a level more amenable to testing. The process does not create new information, but only replicates already-existing DNA information. See State v. Parker, 350 S.W.3d 883, 894 (Tenn. 2011). During this step, particular areas of DNA are also marked with fluorescent dyes in an automated process. Williams, 132 S. Ct. at 2253-54 (Breyer, J., concurring). Once more, the analyst’s role is to facilitate the operation of a machine, not to make any factual affirmation and not to serve as a “witness against” anyone.
¶64 Fourth, once the DNA has been replicated and amplified, it is processed by a capillary electrophoresis instrument. The machine passes the DNA through a gel matrix that filters particles by size, slowing down larger particles more than smaller ones. With the particles filtered by the gel matrix, a light beam can capture the fluorescent markings on the DNA and identify peaks and repeats in the makeup of the DNA. This step is fully automated; like the gas chromatograph that Justice Sotomayor discussed in the concurrence in Bullcoming, it is not testimonial and does not invoke the confrontation clause. 131 S. Ct. at 2723.
¶65 Finally, the machine outputs an electropherogram, or a plot of the peaks and valleys in the DNA. Only here *488does any element of human decision-making enter the process; an expert must translate the peaks and valleys of the electropherogram into a DNA profile. The expert can then prepare an allele table that compares the DNA profiles taken from various sources. See State Ex. 136, at 4. Unlike previous steps in the process, the DNA profile is an affirmation of fact; it is a conclusory statement that a given DNA donor has certain genetic characteristics. But this does not necessarily make the DNA profile into a confrontation clause matter, for two reasons. First, the DNA profile was not introduced into evidence, but was shown only during Pineda’s testimony for illustrative purposes. It is hard to imagine that a chart that never enters into evidence is meant as “an out-of-court substitute for trial testimony.” Bryant, 131 S. Ct. at 1155. And the second component of the confrontation clause must still be satisfied — the statement must still be “against” someone.
2. Application of Confrontation Clause Test
¶66 The necessary inculpatory element enters the equation once an expert compares the DNA profiles. Done on the stand in open court, this comparison is permissible as a “straightforward application of [the analyst]’s expertise.” Williams, 132 S. Ct. at 2270 (Kagan, J., dissenting). The confrontation clause is implicated only at the comparison stage because an allele table does not itself identify (let alone inculpate) anyone, nor would it have any particular meaning to a nonexpert. As the State described the allele table, it appears as “a whole bunch of numbers that kind of look like gobbledygook.” 12 RP at 1538. Pineda’s expertise was necessary to explain what the numbers represented (the number of times each of 16 specified gene sequences repeated) and why they were significant (the DNA of James Anthony Negron, an alternate suspect, failed to match DNA recovered from the crime scene). In other words, the allele table lacked the inculpatory character of the certificate *489reading “cocaine” in Melendez-Diaz, the blood alcohol concentration report in Bullcoming, or the witness statements in Crawford and Davis. The DNA profiles gained their inculpatory character, thus precipitating the confrontation clause issue, only when Pineda’s testimony inculpated Lui.
¶67 Accordingly, the only “witness against” the defendant in the course of the DNA testing process is the final analyst who examines the machine-generated data, creates a DNA profile, and makes a determination that the defendant’s profile matches some other profile. Absent that expert analysis, we are left with an abstract graph or set of numbers that has no bearing on the trial. Pineda was not a surrogate witness whose only purpose was to act as a channel for the DNA profile to enter into evidence. If she was, then the prosecution would hardly be served, or the defendant identified, by a page of meaningless “gobbledygook.” 12 RP at 1538. Rather, Pineda examined the electropherogram and made the determination that Lui could not be excluded as a possible DNA donor.10 Pineda produced her own analysis, “an original product that can be tested through cross-examination.” United States v. Johnson, 587 F.3d 625, 635 (4th Cir. 2009).
¶68 The dissent argues that the jury saw reports prepared by other case analysts who were never subject to cross-examination. Dissent at 512-13. It relies on Pineda’s testimony that Hunan Nasir interpreted the results of samples and wrote the reports. 12 RP at 1552. The dissent ignores Pineda’s testimony that she prepared the exhibit to use with her testimony. Id. at 1497-98; see State Ex. 136 (admitted for illustrative purposes only and did not go to the jury room). Pineda also testified that she came to her own results. 12 RP at 1507. She did not simply rely on the conclusions made by Nasir. Id. She looked at the electronic *490data from the samples, drew her own interpretations, and reached her own conclusion. Id. Nothing in the record states that the jury saw the reports prepared by Nasir. In fact, the exact opposite is true: Pineda testified that she prepared the exhibit. Id. at 1497-98.
¶69 In looking to the ultimate expert analysis, and not the lab work that leads into that analysis, we follow the Court in distinguishing between a person who attests to some fact and a person who aids an expert witness in reaching an attestation of fact: Melendez-Diaz stressed that live testimony is not required if it merely helps to establish “the chain of custody, authenticity of the sample, or accuracy of the testing device.” 557 U.S. at 311 n.l. This category of evidence encompasses taking the clippings, adding the chemicals, and running the machines. It was Pineda who took the results from the capillary electrophoresis machine and reached a conclusion of fact from them (thus becoming a witness) and who testified that the conclusion of fact weighed on an issue in Lui’s case (thus becoming a witness against Lui).
¶70 While Pineda did not personally observe the lab tests that underlaid her analysis — that is, the first four steps of the DNA testing process— Bullcoming guarantees the accused the right “to be confronted with the analyst who made the certification,” 131 S. Ct. at 2710, and not the analysts whose work might have contributed to that certification. Even if Bullcoming required the testifying witness to have personal knowledge of the forensic testing, Pineda had such knowledge. Pineda cannot be analogized to the surrogate analyst in Bullcoming, or even to the expert in Williams, who did not work for Cellmark, had no knowledge of Cellmark’s operations, and “for all the record discloses ... may never have set foot in Cellmark’s laboratory.” 132 S. Ct. at 2268 (Kagan, J., dissenting). Rather, Pineda was an experienced supervisor with Orchid and was well informed about the procedures used and observations made. She reviewed the results of the control samples, she *491reviewed the testing procedures, and she reviewed her subordinate analysts’ results at each step in the process. She was “a supervisor, reviewer, or someone else with a personal, albeit limited, connection to the scientific test at issue.” Bullcoming, 131 S. Ct. at 2722 (Sotomayor, J., concurring).
¶71 The problem in Melendez-Diaz and Bullcoming was that the defendant was denied effective cross-examination — in Melendez-Diaz because the witness was absent, 557 U.S. at 308, and in Bullcoming because the witness lacked the relevant knowledge, 131 S. Ct. at 2707. But here, Lui had the opportunity to cross-examine Pineda on how she arrived at her interpretation and conclusion from the electropherogram. He did not do so, instead focusing his cross-examination on the imprecision inherent to the procedure Orchid used: “Y-STR” testing focuses on the Y chromosome, which is the same in every person in the same paternal lineage, meaning that a DNA match for Lui is also a DNA match for Lui’s son, father, grandfather, and so on. Lui was given a meaningful opportunity to impugn Orchid’s DNA evidence, and he in fact did so to the best of his ability.
¶72 It is unclear how Lui’s confrontation clause rights would be further vindicated by requiring the State to call “anyone whose testimony may be relevant in establishing the chain of custody, authenticity of the sample, or accuracy of the testing device ..., ” and the Supreme Court has never imposed such a requirement. Melendez-Diaz, 557 U.S. at 311 n.l. Indeed, such an onerous requirement would, “for all practical purposes, forbid [ ] the use of scientific tests in criminal trials.” Id. at 332-33 (Kennedy, J., dissenting). Empirical data show that defendants frequently demand to confront scientific witnesses: since Melendez-Diaz was decided, some state laboratories have estimated that their backlog of drug tests has risen by 40 percent and their backlog of alcohol and toxicology tests has risen by 15 percent, resulting in delays or dismissals when lab technicians are unable to attend trials. Hon. G. Ross Anderson Jr., *492Returning to Confrontation Clause Sanity: The Supreme Court (Finally) Retreats From Melendez-Diaz and Bull-coming, 60 Fed. Law. 67, 71 (2013). In addition, an analyst may be unable to attend trial due to sickness, travel, inclement weather, or being called to testify in another trial. Melendez-Diaz, 557 U.S. at 343 (Kennedy, J., dissenting). An analyst may also die prior to trial, an issue not yet resolved by Supreme Court precedent. Williams, 132 S. Ct. at 2251 (Breyer, J., concurring).
¶73 The rule we adopt today avoids the risk of unduly burdening the use of scientific evidence, while preserving the benefits of using a multiplicity of analysts. In the context of DNA testing, as many as 12 different analysts may be involved in a single case. Id. at 2252 (Breyer, J., concurring). Spreading out the testing procedure among different analysts not only promotes “efficiency in the laboratory,” 12 RP at 1572, it also is an important safeguard against fabrication. A laboratory that uses multiple analysts can segregate the analysts working on the suspect’s sample from the analysts working on the crime-scene sample. Williams, 132 S. Ct. at 2252 (Breyer, J., concurring). Screening off the analysts from each other frustrates the potential fabricator, who will have access to only one of the two DNA profiles needed for a match to exist. Furthermore, the use of multiple analysts is also a useful safeguard against error because multiple analysts can review each other’s work in a way that a single analyst cannot.
¶74 In short, we decline to adopt a rule that encourages laboratories to entrust cases to lone analysts, thereby giving the hypothetical rogue analyst more opportunities to fabricate results. More generally, we decline to adopt a rule that makes DNA evidence unduly burdensome to introduce because the use of less reliable eyewitness evidence may increase the risk of erroneous convictions. Id. at 2251 (Breyer, J., concurring). Neither result is helpful to defendants or to the State, and neither result furthers the confrontation clause’s goals of checking potential error or *493fabrication. Under our test, Pineda was the only person involved in Orchid’s testing process who made a statement of fact that tended to inculpate Lui. Thus Pineda was exactly the right analyst to satisfy the strictures of the confrontation clause. The State properly introduced the DNA evidence through her.
D. Temperature Readings
¶75 Lui also asserts that the State was required to introduce the temperature readings of Boussiacos’s body and the ambient temperature through the live testimony of Dr. Kathy Raven, who took those readings. But like the Orchid DNA analysts, Raven did not become a “witness against” Lui by merely taking temperature readings. She may have been a “witness” by virtue of recording the temperatures, thus creating factual information for later use by the court. But she did not do so “against” Lui. Like the raw DNA profile, Raven’s temperature data had no relevance to Lui’s case until Harruff used that data to estimate a range for the time of death. In fact, until Harruff determined that the time of death was consistent with the prosecution’s theory, the temperature readings could well have benefited Lui. In other words, the first prong of our test was met — Raven created a factual statement of the temperature of Boussiacos’s body and of the ambient environment. But the second prong was not met because these points of data could not inculpate Lui without the intervening analysis of an expert. Because Harruff used his expertise to turn raw data into a conclusion that inculpated Lui, it is Harruff and not Raven with whom the confrontation clause is concerned.
E. Toxicology and Autopsy Reports
¶76 Lui objects to the admission of evidence that Boussiacos had no drugs in her system at the time of death. He asserts that Dr. Richard Harruff had no personal *494knowledge about the toxicology screen on Boussiacos’s body and did not offer his professional opinion about the results; rather, he argues that Harruff was merely a mouthpiece for the conclusions of an absent analyst. Similarly, Harruff testified to statements taken directly from the autopsy report about which he had no personal knowledge. These include that Boussiacos had bruising under her scalp and within her neck muscles, that her hair was in a ponytail, that she was wearing a long sleeved shirt that was pulled up toward her chest, that her bra was wadded up and placed underneath her shirt, and that the examining pathologist described certain scratches as “contusions.”
¶77 Lui is correct. Like the temperature readings, the information taken from the toxicology report and autopsy report were statements of fact. But unlike the temperature readings, these statements had an inculpatory effect: the toxicology report was prepared to identify the cause and manner of Boussiacos’s death and was relied on at trial to rebut Lui’s testimony that Boussiacos might have been smoking prior to her death. The statements from the autopsy report were also for the purpose of identifying the manner of death and were used to prove that Boussiacos was dressed postmortem. All of the statements were used by the prosecution to convict Lui. Furthermore, unlike Harruff’s testimony based on the autopsy photographs or temperature readings, Harruff did not bring his expertise to bear on the statements or add original analysis — he merely recited a conclusion prepared by nontestifying experts.11 His testimony falls precisely within the scope of testimony proscribed by Bullcoming, and it was error to *495admit the toxicology report and statements from the autopsy report.
¶78 However, these erroneous admissions do not warrant a new trial. A constitutional error is harmless if the appellate court is assured beyond a reasonable doubt that the jury verdict cannot be attributed to the error. State v. Watt, 160 Wn.2d 626, 635, 160 P.3d 640 (2007). Under our “ ‘overwhelming untainted evidence’ ” test, we look to the untainted evidence to determine if it was so overwhelming that it necessarily leads to a finding of guilt. State v. Guloy, 104 Wn.2d 412, 426, 705 P.2d 1182 (1985).
¶79 We applied the “overwhelming untainted evidence” test in State v. Anderson, 171 Wn.2d 764, 254 P.3d 815 (2011). There, the defendant was charged with first degree child molestation for allegedly abusing a child referred to as M.A.E. Id. at 766. The State introduced the testimony of another alleged victim of Anderson’s, C.C.S., through a nurse practitioner who had examined him. The State also introduced Anderson’s confession to molesting a third child, K.R.P.; the live testimony of the detective who had investigated KR.P.’s case; and the testimony of M.A.E. himself. Id. Anderson appealed, arguing that the trial court had erred by permitting the nurse to testify to C.C.S.’s statements. We held that even if the introduction of C.C.S.’s statements violated the confrontation clause, any error was harmless. Id. at 770. We noted that aside from C.C.S.’s statements, there was also unchallenged testimony by M.A.E. and evidence of Anderson’s prior abuse of K.R.P. In comparison, we described the nurse’s testimony as “cursory” and pointed out that it occupied less than one page of the trial transcript. Id. For those reasons, we held that “[g]iven the amount of evidence presented and the fact that nurse Young’s testimony added little, if any, evidence to prove the elements of the current charge against Anderson, we find that the trial court’s error in admitting nurse Young’s testimonial statements was harmless.” Id.
*496¶80 This case is much the same. Like the testimony in Anderson, the toxicology report does nothing to prove any element of the charges against Lui. Boussiacos’s toxicology was implicated only to rebut Lui’s statement to police that Boussiacos might have been murdered while sneaking out to smoke. Lui’s false statement to investigators, however, was only one of many reasons for the jury to doubt Lui’s credibility. As the prosecution showed, Lui repeatedly changed his accounts of when Boussiacos planned to depart for California, when she packed for her trip, and when he became aware of Boussiacos’s intention to leave. Lui’s testimony about his affair, his phone call with his sister at 1:00 a.m., Boussiacos’s favorite pajamas, and Boussiacos’s engagement ring was flatly inconsistent with the testimony of other witnesses. Finally, DNA evidence contradicted Lui’s denial that he had had sexual intercourse with Boussiacos prior to her death. Aside from the toxicology report, the State produced significant evidence detracting from Lui’s credibility.
¶81 The statements taken from the autopsy report about Boussiacos’s injuries were minor compared to the properly admitted evidence. Harruff, trained in forensic pathology, examined photographs taken both at the crime scene and during the autopsy. Harruff describes the injuries in the photographs in over 20 pages of direct examination. Harruff testified that the photographs showed a number of prominent abrasions on Boussiacos’s neck and petechiae12 on her face, indicating strangulation and a struggle. There were also a number of other injuries, including bruising on her face, shoulders, left hand, and armpit region. Harruff testified that the color of the abrasions indicated that the injuries occurred close to the time of death.
¶82 The other statements taken from the autopsy report were minor. Harruff was asked by the State if page 3 of the autopsy report indicated how Boussiacos’s hair was styled. *497He responded that her hair was pulled back in a ponytail. However, the jury later saw a picture of the left side of Boussiacos’s hair. The jury also saw photographs of Boussiacos’s pants, underwear, shoes, and socks, indicating that she was abnormally dressed. Additionally, her shirt and shoes were present in court. Harruff’s improper testimony was minor compared to the evidence properly admitted. And, properly admitted evidence corroborated most of his statements.
¶83 Significant other untainted evidence supported a finding of guilt. The State produced evidence showing Lui was angry that Boussiacos was going to “leave him and reclaim her own life,” 14 RP at 1805; evidence suggesting that Boussiacos had died before she could dress or put on her customary makeup; evidence suggesting that Boussiacos had been dressed and her bags packed by “somebody who doesn’t know anything about women,” 14 RP at 1838; testimony that a police bloodhound led investigators directly to Lui; evidence of Lui’s familiarity with the area where Boussiacos was found; and evidence of Lui’s unusual conduct during the investigation. The weight of this evidence makes it implausible that the jury’s verdict could be attributed to the toxicology and autopsy reports — evidence that was discussed in 2 pages of a nearly 2,000 page record. 10 RP at 1397-98. The cumulative, untainted evidence necessarily led to a finding of guilt, and thus the Court of Appeals correctly rejected Lui’s confrontation clause challenge to the admission of the toxicology and autopsy reports.
CONCLUSION
¶84 The State satisfied the confrontation clause when it produced the “witnesses against” the defendant — that is, analysts who use their expertise to reach a factual conclusion bearing on an issue in Lui’s case. Lui’s claim to the contrary overlooks the plain language of the confrontation *498clause and its overarching purpose of preventing civil-law examinations of the type used in Sir Walter Raleigh’s trial. Crawford, 541 U.S. at 44. The absent DNA analysts here cannot be compared to the absent Lord Cobham in that trial: their role was limited to handling and manipulating evidence and machines, analogous to the chain of custody. While we are sensitive to the risk of erroneous or fraudulent lab conduct, the potential for error is no greater than the risk of mishandling of evidence somewhere along the chain of custody. Because the chain of custody is not a confrontation clause issue, neither is laboratory work, without more.
¶85 Here, Pineda was Lord Cobham to Lui’s Raleigh: by taking the output of a laboratory process and using her expertise to construct an incriminating DNA profile from the data, she became Lui’s accuser. The confrontation clause required Pineda’s testimony, which the State produced. Therefore, there was no error. Neither was there error in the court’s admission of temperature or reversible error in the admission of the toxicology or autopsy evidence. We affirm the Court of Appeals.
Madsen, C.J., and C. Johnson, J.M. Johnson, and González, JJ., concur.U.S. Const, amend. VI; Const, art. I, § 22.
For the sake of simplicity, we analyze Reliagene’s tests and Orchid’s tests together and refer only to Orchid in the following analysis.
U.S. Const, amend. VI.
The dissent suggests that our analysis is precluded by Melendez-Diaz because Justice Thomas signed the majority. Dissent at 500 n.13. That would be true only if formalized evidence, signed by the author, was admitted into evidence. Instead, we have an expert witness relying upon a report that did not come into evidence.
Justice Thomas declined to join note 6 and Part IV, which defined testimony subject to the confrontation clause without including a requirement of indicia of formality. Bullcoming, 131 S. Ct. at 2709.
U.S. Const, amend. VT.
The dissent asserts that we have characterized lab results as “inherently accurate.” Dissent at 515. However, our focus is not on whether the work of analysts is accurate and reliable, but on whether it is “testimony against” an accused.
The dissent fails to grasp this test, which is made clear by its comparison between a police supervisor testifying about eyewitness observations that were recorded by subordinate police officers and an expert witness relying on work done by other lab analysts when arriving at his or her own conclusions. Dissent at 520. The confrontation clause is triggered when an expert witness makes a factual assertion that is used in court and that factual assertion is inculpatory. This test applies only to expert witnesses. The dissent’s comparison fails for two reasons. Id. First, the example fails to specify that the police supervisor is not a conventional witness simply describing a crime scene. If the police officer is a conventional witness, the primary purpose test applies. Second, even if the testimony was from an expert witness, the observations would likely be statements of fact that are inculpatory, triggering the confrontation clause. The statements would not need to be interpreted or analyzed by an expert. They would have meaning on their own, unlike DNA data or the temperature of a body, without being further analyzed.
The dissent cites to note 2 in Crawford for its argument that “seven justices held that the confrontation clause applies to all witnesses against the accused regardless of whether the witnesses are neutral or are experts, such as coroners.” Dissent at 502. The note in Crawford simply states, “[Sjeveral early American authorities flatly rejected any special status for coroner statements.” Crawford, 541 U.S. at 47 n.2. The note discusses the history of the confrontation clause and whether coroner statements were exempt from the clause. See id. Our holding does not create a confrontation clause exception for coroners or for expert witnesses. The dissent also cites to a discussion in Crawford that rejected the argument that the wife’s statements were exempt from the confrontation clause because they were made to a “neutral” government officer. Id. at 66. Again, our opinion does not create an exception to the confrontation clause. We are clarifying when an expert witness may rely on the work of others when reaching his or her own opinion without violating the confrontation clause.
See 12 RP at 1507 (“I came to my own results____[I]n the end, all of the data is reduced to electronic format. Once it comes out of the machine ... it is what we call an electropherogram, or a plot. ... I did look at the electronic data from the results. ... I did draw my own interpretation and my own conclusions from it.”).
The dissent suggests that it is inconsistent that we could find a toxicology report to be inculpatory but not the DNA testing. Dissent at 507. However, DNÁ testing becomes significant only when data is turned into a profile and that profile is compared to known samples. It is at this point that the data becomes meaningful to a case and inculpatory. In comparison, the toxicology report was an inculpatory statement without being interpreted by Harruff.
“Petechiae” are small red dots on the skin resulting from ruptured blood vessels.